© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 14847
    Prev
    Next
    INDEX
    Explanations

    alternative solutions or viewpoints compared to the mainstream or expected ones

    oai_token-act-pair · gpt-3.5-turbo
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    288
    +0.13
    0.6%
    276
    +0.10
    0.5%
    754
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    276
    +0.13
    0.02
    221
    +0.10
    0.02
    754
    +0.10
    0.02
    Negative Logits
    older
    -0.84
    hips
    -0.77
    Chicken
    -0.72
    ursed
    -0.72
    yers
    -0.70
    ching
    -0.70
    oho
    -0.70
    gin
    -0.69
    uchi
    -0.69
    geries
    -0.68
    POSITIVE LOGITS
     lifestyles
    0.90
     viewpoints
    0.85
     explanations
    0.79
    atives
    0.78
     therapies
    0.77
     thereto
    0.76
     solutions
    0.76
     Altern
    0.76
     alternatives
    0.75
     modes
    0.75
    Activations Density 0.028%

    No Known Activations